Nucleoprotein-RNA Orientation in the Measles Virus Nucleocapsid by Three-Dimensional Electron Microscopy

Abstract

Recombinant measles virus nucleoprotein-RNA (N-RNA) helices were analyzed by negative-stain electron microscopy. Three-dimensional reconstructions of trypsin-digested and intact nucleocapsids coupled to the docking of the atomic structure of the respiratory syncytial virus (RSV) N-RNA subunit into the electron microscopy density map support a model that places the RNA at the exterior of the helix and the disordered C-terminal domain toward the helix interior, and they suggest the position of the six nucleotides with respect to the measles N protomer.

The RNA genome of nonsegmented negative-strand RNA viruses is tightly and regularly encapsidated by the viral nucleoprotein N, providing flexible helical templates for viral transcription and replication. Upon heterologous expression, nucleoproteins associate not only with long cellular RNAs to form helical nucleocapsids undistinguishable from the viral ones but also with short cellular RNAs that noncovalently close up into N-RNA rings. In the rings, N-RNA is sterically constrained in a biologically inactive form, but the rings have an advantage of being rigid enough for X-ray crystallography. Conversely, the helical assemblies are challenging for electron microscopy (EM) analysis because of their flexibility but are the biologically relevant ones.

The atomic structures of N-RNA rings of rabies virus and vesicular stomatitis virus (both rhabdoviruses) (1, 10) reveal the shielding of RNA between two domains of N in a positively charged cleft situated inside the rings. Extended N- and C-terminal domains reach out to neighboring N protomers in order to stabilize and rigidify the structure. Recently, the structure of N-RNA rings of respiratory syncytial virus (RSV; a paramyxovirus) was determined (24). The global architecture of the nucleoprotein is very similar to that of the rhabdoviruses, although there are 7 ribonucleotides (nt) instead of 9 bound to each N protomer. However, the lateral contacts between adjacent N subunits of the ring confer to it an opposite curvature, which results in an outward RNA groove location. RSV N has an N-terminal exchange domain similar to that of rhabdovirus N, but the C-terminal domain is slightly different, as it is not clearly involved in contacts between subsequent N protomers. Is this inversion of the subunit orientation due simply to steric constraints in the ring, or does it also take place in a helical nucleocapsid? Tawar and coworkers modeled an RSV N-RNA helix but could not directly dock the atomic structure of RSV N into their helical EM reconstruction (24).

A sequence alignment between RSV N and measles virus N (MeV N), both paramyxovirus nucleoproteins, is difficult to interpret because of the lack of amino acid identity. However, a comparison of the secondary structure elements observed in the RSV N structure, with a secondary structure prediction for MeV N (6) (Fig. 1), shows even more similarity than that between rhabdovirus and RSV N. This comparison also shows that the β-hairpin projecting from the distal end of the RSV N protomer (24) is conserved between these two paramyxovirus nucleoproteins. One important difference lies in the length of the highly disordered C-terminal domain, the N tail, that is 31 residues long (360 to 391) for RSV N (24) but 126 residues long (400 to 525) for MeV N (16). A short sequence in the MeV N tail (residues 489 to 506) folds into a dynamic helical structure that is stabilized by binding of the viral phosphoprotein that carries the viral RNA-dependent RNA polymerase (12, 13, 16). The N tail is also involved in binding host proteins, such as hsp70 (5, 26) and interferon regulatory factor 3 (14, 15). So far, the location of the N tail on the helix is not known, although it is usually shown on the outside in cartoons that illustrate transcription and replication of paramyxoviruses (see Fig. 9 in reference 3). The helical model derived from the recombinant N-RNA ring structure of RSV, however, would place the N tail toward the helix interior, which would have consequences for the contacts between subsequent helical turns.

FIG. 1.

Predicted secondary structure of MeV N compared to secondary structure elements in the atomic structure of RSV N. α-Helices are represented as red boxes, β-strands as blue arrows.

The helical structure of the intact measles virus N-RNA under cryoelectron microscopy (cryo-EM) conditions is highly flexible and difficult to determine by Fourier-Bessel image analysis or even by single-particle-based approaches (2, 21). However, once the N tail is removed by proteolysis, the structure becomes more regular and rigid and thus amenable to helical reconstruction by cryo-EM (21). Here, we show that the nondigested nucleocapsid structure can be addressed in negative-stain electron microscopy by trapping the sample between two layers of carbon film and by using NanoW stain (from Nanoprobes) instead of the more traditional uranyl acetate (Fig. 2). This preparation technique enables to image intact measles virus nucleocapsids as well as their trypsin-digested counterparts and has the advantage of maintaining the helix in a more rigid state. For this analysis, recombinant MeV N was produced, and a fraction of it was trypsinated as described previously (21) and imaged with a transmission electron microscope. Overlapping segments of the visually most rigid helices were selected with Boxer (17), contrast transfer function (CTF) corrected with CTFFIND3 (18) and Bsoft (11), and aligned and classified with Imagic (25). An additional classification of power spectra of individual image frames and a sorting based on artificial smooth helical volumes improved the homogeneity of different subsets separated according to diameter and helical parameters. The major subsets were used for angular assignment and three-dimensional (3D) reconstruction in an iterative projection-matching procedure similar to that for IHRSR (7, 8) with the SPIDER package (9, 22), starting from a smooth helix of a chosen pitch as the initial model (for details, see the supplemental material).

FIG. 2.

Fields of view of negatively stained MeV nucleocapsids. (A) Intact nucleocapsids with 2% uranyl acetate and a single carbon layer. (B and C) Intact (B) or digested (C) nucleocapsids with NanoW in a double carbon layer and a representative class average of power spectra. (D) Recombinant C-terminally His-tagged nucleocapsids bound to anti-His-tagged antibody.

Final three-dimensional reconstructions of trypsin-digested and intact measles nucleocapsids at a resolution of 25 Å are shown in Fig. 3. Removal of the N tail leads to a compaction of the helix, with the pitch shortening from 57.2 Å to 48.7 Å and a diameter constriction from 200 Å to 190 Å, in line with the previous cryo-negative-stain EM work (2). The number of subunits per turn in the digested nucleocapsid was found to be 12.33, the same as that previously obtained for such species under cryo-EM conditions by Fourier-Bessel analysis of the most regular helix coupled to IHRSR (21). Thus, in this case, the double-carbon layer negative-stain microscopy and the NanoW stain seem to maintain the helical structure without modifying the helical parameters. The intact nucleocapsid helix accommodates a nearly integer number of 12.92 subunits per turn, which agrees with the 5% increase in diameter. The overall shape of the nucleoprotein subunit is nevertheless very similar in both reconstructions, corroborating previous arguments in favor of the intrinsic N-tail disorder (2, 16).

FIG. 3.

Three-dimensional reconstructions of MeV nucleocapsids. (A and D) Digested nucleocapsid, (B and E) intact nucleocapsid, (F) cryo-EM reconstruction of digested MeV N (21). The fit of the RSV N-RNA atomic structure is shown as an overlay. The N-terminal β-hairpin fits nicely into the density (arrow). The RNA is shown as a red ribbon. (C) Docking precision for panel A. Shown is the correlation upon rotation of the monomer (see the supplemental material).

Given the predicted structural similarity between RSV and MeV N, the atomic model of the RSV nucleoprotein monomer (Protein Data Bank [PDB]accession number 2WJ8) was used for fitting into the obtained 3D volumes with VEDA (http://mem.ibs.fr/VEDA), a new graphical version of URO (19) (Fig. 3). For fitting, MeV nucleoprotein helices were considered to be left-handed based on previously published metal shadowing results (21), and a modified PDB file of the RSV N protomer with only 5 nt corresponding to nt 2 to 6 was used, given that MeV N-RNA contains 6 nt per N protomer (4, 23) and not 7. Interestingly, without any constraints imposed during fitting, the fit ensures the continuity of RNA bound to measles virus N. Atomic coordinates of two RNA segments bound to consecutive subunits were extracted from the thus-obtained MeV nucleocapsid model, an additional ribonucleotide (corresponding to number 7 in Fig. 1D in reference 24) was inserted, and energy minimization was performed with VEGA software (20) to obtain a continuous, physically realistic RNA molecule. Since bases 2, 3, and 4 bind in a cavity on the RSV-N protein, their coordinates were kept fixed, while those of the solvent-facing ribonucleotides, 5, 6, and 1, were optimized. Figure 4 illustrates the possibility of easily constructing a 6-nt RNA with three bases facing the protein, as in the RSV N-RNA rings, and three bases stacked and pointing away from the protein into the solvent.

FIG. 4.

RNA binding to MeV N based on RSV N-protomer fitting and energy minimization for solvent-exposed bases. Protein-oriented bases are in green, solvent-oriented bases are in blue, and the backbone is in light blue. (A) Enlarged view of RNA binding. (B) Schematic diagram for a comparison of RNA interaction with MeV N and RSV N. The numbering is as in reference 24. The gray nucleotides are on the neighboring N protomers. (C) Top view of the helical fit.

This fit of the atomic structure of RSV N into the negative-stain EM reconstructions is also consistent with the previously published cryo-EM structure of the digested MeV nucleocapsid (21) (Fig. 3F) and the RNA position predicted therein by cis-platinum RNA labeling. It suggests that the RNA is indeed localized at the exterior face of the helix, as in the RSV N-RNA rings, and not as in rhabdoviral N-RNA rings. Although the disordered C-terminal domain could not be resolved in the reconstruction of the intact nucleocapsid, the fit suggests that this crucial domain would point toward the helix interior. In addition, binding of anti-His-tagged antibody to C-terminally His-tagged nucleocapsids prevents correct helix formation (Fig. 2D) (see the supplemental material), indicating that the N tail domain may come out at a site where it interferes with contacts between two subsequent turns of the N-RNA helix, contributing to its flexibility.